Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=wcis20

Download by: [Tufts University] Date: 19 September 2016, At: 12:21

Computers in the SchoolsInterdisciplinary Journal of Practice, Theory, and Applied Research

ISSN: 0738-0569 (Print) 1528-7033 (Online) Journal homepage: http://www.tandfonline.com/loi/wcis20

Programming with the KIBO Robotics Kit inPreschool Classrooms

Mollie Elkin, Amanda Sullivan & Marina Umaschi Bers

To cite this article: Mollie Elkin, Amanda Sullivan & Marina Umaschi Bers (2016) Programmingwith the KIBO Robotics Kit in Preschool Classrooms, Computers in the Schools, 33:3, 169-186,DOI: 10.1080/07380569.2016.1216251

To link to this article: http://dx.doi.org/10.1080/07380569.2016.1216251

Published online: 15 Sep 2016.

Submit your article to this journal

View related articles

View Crossmark data

COMPUTERS IN THE SCHOOLS, VOL. , NO. , –http://dx.doi.org/./..

Programming with the KIBO Robotics Kit in PreschoolClassrooms

Mollie Elkin, Amanda Sullivan, and Marina Umaschi Bers

Tufts University, Medford, Massachusetts, USA

KEYWORDSEarly childhood education;preschool; programming;robotics

ABSTRACTKIBO is a developmentally appropriate robotics kit for young chil-dren that is programmed using interlocking wooden blocks; noscreens or keyboards are required. This study describes a pilotKIBO robotics curriculum at an urban public preschool in RhodeIsland and presents data collected on children’s knowledge offoundational programming concepts after completing the cur-riculum. The curriculum was designed to integrate music, liter-acy, and design with engineering and robotics. Children (N = 64)from seven preschool classrooms, ranging in age from 3 to 5, par-ticipated in the study. Findings indicated that children as youngas age 3 could create syntactically correct programs for the KIBOrobot, although older preschoolers (closer to age 5) performedbetter than younger preschoolers on a standardized program-ming task. Additionally, all students generally performed betteron the programming tasks that required them to manipulate lessprogramming instructions. Implications for designing develop-mentally appropriate curriculum and scaffolding for young chil-dren are addressed.

New technologies are increasingly influencing the ways young children are grow-ing, learning, and playing. Digital activities such as playing video games and usingan iPad are growing in prevalence among young children under the age of 8. Forexample, a recent study by Common Sense Media (2013) found that two thirds ofchildren ages 0 to 8 have access to a console video game player at home, and 35%have access to a handheld game player such as a Game Boy, PSP, or Nintendo DS.Additionally, there has been a fivefold increase in ownership of tablet devices suchas iPads from 8% of all families in 2011 to 40% in 2013.

As technology has grown increasingly common in young children’s home envi-ronments in recent years, educational technology in schools has also expanded. Thishas occurred in part due to federal education programs and private initiatives mak-ing computer science and technological literacy a priority for young children (Officeof Educational Technology, 2010). Robotics and computer programming initiatives

CONTACT Mollie Elkin elkin.mollie@gmail.com DevTech Research Group, Tufts University, College Avenue,Medford, MA .Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/wcis.© Taylor & Francis Group, LLC

170 M. ELKIN ET AL.

for young children have grown in popularity over the past 5 years as new productsfor young learners have emerged on the commercial market (i.e., KIBO, Bee-bot,Dot, and Dash).

Prior research has shown that children as young as age 4 can successfully buildand program a simple robot (Bers, Ponte, Juelich, Viera, & Schenker, 2002; Cejka,Rogers, & Portsmore., 2006; Perlman, 1976; Sullivan & Bers, 2015; Sullivan, Kaza-koff, & Bers, 2013; Wyeth, 2008), but there is still very limited research on whatchildren under age 4 can learn with robotics. Sullivan and Bers (2015) found thatpreschool students were able to successfully complete basic programming tasksupon completion of a Kids Invent With Imagination (KIWI) robotics curriculum.Similarly, Sullivan et al. (2013) found that with scaffolding, 5-year-old preschoolchildren were able to design, build, and program a LEGO® WeDo robot. Both ofthese studies emphasized the importance of scaffolding and moving through thecurriculum at a slower pace than when working with children in kindergartenthrough second grade. The present study builds on prior research with preschoolersby looking at what children as young as age 3 can learn about foundational program-ming concepts and skills when completing a developmentally appropriate curricu-lum that includes built-in scaffolding and review time. This article presents resultsfrom a pilot experience in a preschool robotics program at a public school in RhodeIsland.

Literature review

Robotics in early education

From tablet devices to newly designed robotics kits, young children are exploringdifferent types of technology while at school. While access to new technologies isgrowing, children’s understanding of how and why these tools work the way theydo is growing as a new area of research. Robotics and computer programmingoffer a way to playfully engage students with the process of how motors, sensors,and electronics work (much the way they work in automated doors, sinks, anddigital toys) through hands-on building projects (Bers, 2008). Young children arenaturally inquisitive about how things work and are willing to take risks to uncoversolutions (Peel & Prinsloo, 2001). Robotics offers an environment for children totest their hypotheses, engage in problem solving, and make personally meaningfuldiscoveries.

In recent years, there has been a range of new robotic kits on themarket for youngchildren. For example, Bee-Bot can be used to teach sequencing, estimation, andproblem solving. Children program it by pushing on the directional buttons locatedon the robot’s body (www.be-bot.us).More recently, themakers of Bee-Bot have cre-ated Blue-Bot, which functions similarly to the Bee-Bot but can also be programedfrom a tablet or computer; the program is sent to the robot through a Bluetoothconnection (www.terrapinlogo.com/robots.html). Another example is the Dash andDot robots, created by Wonder Workshop. Children program these robots throughiPad and Android applications to navigate a route, as well as use lights and sensors.

COMPUTERS IN THE SCHOOLS 171

Some of the applications target children of all ages, whereas others target childrenolder than age 8 (www.makewonder.com). For the research presented in this study,we have chosen to use the KIBO robotics kit described in the next section.

Research with robotics in early childhood settings has shown that beginning inpreschool, children can learn fundamental programming concepts of sequencing,logical ordering, cause-and-effect relationships, and engineering design skills (Bers,2008; Fessakis, Gouli, &Mavroudi, 2013; Kazakoff & Bers, 2011; Kazakoff, Sullivan,& Bers, 2013). When children create programs for their robots, they are sequenc-ing commands for their robot to act out. The act of sequencing is foundational forearlymath, literacy, and planning (Zelazo, Carter, Reznick, & Frye, 1997). Addition-ally, educational robotics programs, when based in research, child development the-ory, and developmentally appropriate practices (National Association for the Edu-cation of Young Children [NAEYC] & Fred Rogers Center, 2012), can foster studentlearning of engineering such as design skills and methods (Druin & Hendler, 2000)while engaging in collaboration and other social skills necessary for school success(Clements, 1999; Lee, Sullivan, & Bers, 2013; Svensson, 2000).

The KIBO robotics kit

Although there are now many commercially available robotic kits that teach aboutprogramming, the majority described in the previous section are “pre-built” in thesense that children are not involved in any of the construction or design aspects ofbuilding a robot. For example, the Bee-Bot robot is designed to resemble a brightand colorful bumblebee, with all motors and design features ready to use, muchlike any children’s toy. In contrast, this study utilized the KIBO robotics kit, whichengages young children in both building and programming. This kit was devel-oped by the DevTech Research Group at Tufts University and commercialized byKinderLab Robotics. KIBO is designed for young children ages 4 to 7 to learn foun-dational engineering and programming content; however, this study examined thehypothesis that it may be developmentally appropriate to use with children as youngas 3 years old. KIBO was chosen for this study because of the large and easy-to-manipulate parts, open-ended building and programming possibilities, and the kit’stangible programming language (Sullivan, Elkin, & Bers, 2015). Because KIBO isprogrammed by putting together wooden blocks, without a computer, tablet, orother form of “screen-time,” curricula utilizing the KIBO kit is aligned with theAmerican Academy of Pediatrics’ (2003) recommendation that young children havea limited amount of screen time per day.

The KIBO kit contains easy-to-connect robotics materials including wheels,motors, light output, and a variety of sensors (see Figure 1). In addition to theseelectronic components, the KIBO kit also contains art platforms that can be usedfor children to decorate and personalize (Sullivan et al., 2015).

KIBO is programmed by using interlocking wooden programming blocks. Thesewooden blocks contain no embedded electronics or digital components, but eachone has a unique barcode. A scanner embedded in the front of the KIBO robotallows users to scan the barcodes on the programming blocks and send a program to

172 M. ELKIN ET AL.

Figure . The KIBO robot and the blocks.

their robot instantaneously (Sullivan et al., 2015). Similar to other programming lan-guages, KIBO has specific syntax rules to follow (see Figures 2 and 3). For example,every program must start with a Begin block and finish with an End block. Addi-tionally, in order to create a functional repeat loop, one must use the Repeat block,a parameter (either a number or sensor), and the End Repeat block.

KIBO’s developmental considerations

Young children’s working memory changes drastically between the ages of 3 and 5(Shonkoff, Duncan, Fisher, Magnuson, & Raver, 2011), enabling them to effectivelylearn new content.When children are entering preschool around age 3,most of themcan organize themselves to complete tasks that involve following two steps, such asthrowing away a napkin and putting away their lunchbox after snack time (RhodeIsland Department of Education [RIDE], 2013; Shonkoff et al., 2011). By the time

Figure . Basic program starting with Begin block and finishing with End block.

COMPUTERS IN THE SCHOOLS 173

Figure . Repeat loops program. Note. The program above shows a syntactically correct repeat loopsprogram using the number parameter. This program will tell the robot to beep three times. Pic-tured below the program are choices of other number parameters that can be used for this type ofprogram.

children are leaving preschool and entering kindergarten around age 5, childrencan follow multi-step instructions and retell familiar stories in the correct sequence(RIDE, 2013). Using the KIBO robot, children can strengthen their working mem-ory skills by learning to sequence increasingly complex programs and to master allof KIBO’s syntax rules.

By building with the robotics manipulatives such as KIBO’s motors, sensors, out-puts, and wooden programming blocks, children are able to develop fine motorskills and hand-eye coordination. Play that involves the manipulation of physi-cal objects with symbolic meaning (i.e., KIBO’s programming blocks that symbol-ize robotic actions) lets children begin to explore more complex symbolic think-ing (Bers, 2008; Piaget, 1952). In addition to these technical manipulatives, chil-dren also exercise their fine motor skills through the addition of arts, crafts, andrecyclable materials. Specifically, the two art platforms provide a space for explor-ing the engineering design process to build sturdy creations that are personallymeaningful (Sullivan et al., 2015; see Figures 4 and 5). The following sectiondescribes the present research evaluating the use of KIBO robotics in a preschoolcontext.

Robotics also engages young children in collaboration and teamwork (Lee et al.,2013). Preschool children are in the developmental process of learning social skillssuch as how to work with others; the design features of certain types of technologycan promote social and prosocial development (Bers, 2012). Unlike many appli-cations and educational software designed for one child working independently,robotics activities lend themselves tomore collaborative moments. For example, theKIBO robotics kit used in this study is designed so that small groups of children canwork on one robot with each one taking on a very specific role: the programmer, theartist, or the engineer.

174 M. ELKIN ET AL.

Figure . KIBO’s static art platform for personalizing projects.

Method

Research overview

This study explores the following research questions:1. What can young preschool children, ages 3 to 5, learn about foundational

programming and robotics content through a short-term educational inter-vention?

2. What types of errors do young preschoolers make when programming withKIBO?

3. What kinds of programming concepts are easiest for young children to mas-ter? Which are more challenging?

To answer these questions, preschool students participated in a nine-hour intro-ductory robotics and programming curriculum. Upon completion of the curricu-lum, students completed aKIBOprogramming task (called “Solve-It”) to assess theirprogramming knowledge.

Figure . The motorized turntable for personalizing projects.

COMPUTERS IN THE SCHOOLS 175

Participants

Participants in this study were 64 predominantly low-income, Hispanic childrendrawn from seven classrooms in an urban public preschool in Rhode Island. Theyranged in age from 3 to 5 years (mean age = 4.83 at the time of assessment). Whilestudents in all of the classes participated in the curriculum, they were not requiredto partake in the assessment portion of the study; all children were invited to com-plete the assessment, but if they did not want to, they could choose to do somethingelse. Students who attend this school are 81% Hispanic, 8% Caucasian, 6% AfricanAmerican, 5% mixed race, and 1% Native American. Additionally, 91% of the stu-dents were eligible for subsidized lunch.

Procedure

Over the course of 3 months, seven preschool classrooms completed an introduc-tory robotics and programming curriculum taught by students from Tufts Univer-sity. These undergraduate and graduate students came from a variety of educationalbackgrounds ranging from the liberal arts to engineering. While some students hadno previous experience teaching robotics, all students had experience working withchildren. Student volunteerswere required to attend two trainings: one 8-hour train-ing before the start of the intervention, and one 4-hour trainingmidway through theintervention in order to practice the curriculum and prepare for administering theassessments. Each volunteer practiced administering the assessment on other vol-unteers before administering the assessment on the children. The children’s regularpreschool teachers were in the classroom at all times to facilitate behavioral man-agement and assist with small group work. A larger goal for the robotics works atthis site was to have classroom teachers learn by observing trained students so theywould be able to implement their own robotics curriculum in the future withoutoutside help.

Curriculum overview

The introductory robotics curriculum involved approximately nine hours of workover the course of 6 days. Each day’s lesson was divided into two parts: 45 minuteswas spent doing an activity with the KIBO robotics kit, and the other 45 minuteswas spent doing robotics and engineering-related activities that did not require theuse of the KIBO robotics kit. Half of the classes completed the robotics portion firstwhile the other completed the non-robotics portion, and then they swapped (sincekits were being shared between classrooms).

During non-robotics time, children spent a portion of the class participating in afull group activity, and then spent the remainder of time participating in an activityof their choice. According to the school’s principal, Rhode Island requires preschoolchildren to spend at least half of their school day engaged in self-directed activities.The KIBO curriculum was therefore designed to include multiple activity choices

176 M. ELKIN ET AL.

Figure . KIBO BINGO.

related to engineering and programming. During the group activity time, childrenlearned songs (such as one that teaches about the different parts of the KIBO robot)and listened to picture books being read aloud (reinforcing fundamental engineer-ing concepts like the engineering design process). During free-choice time, childrencould choose an activity related to KIBO. For example, one activity choice was KIBOBINGO(see Figure 6), which is played like the traditional BINGOgame. The teachershowed a part of the robot (such as a wheel or a body), and students covered up thatpicture on their game board. The goal of this game was to teach children the dif-ferent parts of the robot. Another activity choice was KIBO Says (see Figure 7), an

Figure . Simon Says with programming commands.

COMPUTERS IN THE SCHOOLS 177

Figure . Small group work with robot.

adaptation of the Simon Says game, where children follow the directions on a largeprint-out version of the KIBO blocks if KIBO (the teacher) told them to do so. Inaddition to games, children could choose to create decorations for their robots ordraw in their engineering design journals.

During robotics time, children were given a task to complete involving theirrobot. Each classroom had one robot for approximately three children; the studentvolunteers, as well as the classroom teachers, assisted the different groups. For exam-ple, during Session 2, children were asked to work in small groups to program theirrobots to dance the Hokey Pokey (see Figure 8). The Hokey Pokey involves childrensequencing seven programming blocks in order to make the well-known song. Thisactivity focuses on strengthening young children’s working memory through trialand error and iterative programming (Shonkoff et al., 2011). It also works on theirability to understand sequence and order, which is a foundational early math andliteracy component (Kazakoff & Bers, 2011). During Session 5, children worked insmall groups and programmed their robots to travel along differently shaped pathsusing the Repeat and End Repeat blocks. The repeat loop required children to prac-tice sequencing, order, counting, and estimation to select the correct number param-eter that would make their robot travel the correct distance. See Table 1 for a break-down of the types of activities completed each day.

On the final day of the curriculum, each class was given a KIBO robot kit to buildand program together. Prior to this session, children had learned about differentdances from around the world, and as a class, they selected one dance that theywanted their robot to perform. During robotics time, each class created a danceprogram for their robot. For example, one class programmed their robot to dance theHula, which resulted in a program with the robot repeating the motions of movingleft and right. Another class wanted their robot to move like a Salsa dancer, so theyincluded many Spin blocks in their program. During non-robotics time, studentscreated decorations for the robot itself as well as a “stage” for the robot to danceon. At the end of the curriculum, students presented their dancing robots to special

178 M. ELKIN ET AL.

Table . Overview of the curriculum.

Session Focus Robotics activity Non-Robotics activity

Session Introduction toengineeringand robotics

Discussion about what is a robot,play a game to learn thedifference between a robot, learnthe “KIBO Robot Parts” song

Discussion about what is anengineer, learn the “EngineeringDesign Process” song, completesturdy building activity withnon-robotic materials

Session Introduction towhat is aprogram

Introduction to KIBO’sprogramming blocks, programthe robot to dance the HokeyPokey

Play Simon Says with KIBOcommands, review theengineering design process

Session Introduction tosensing andsensors

Review parts of KIBO robot with the“KIBO Robot Parts” song, reviewthe different blocks with KIBOBingo, free exploration with therobot

Sing “Engineering Design Process”song, read a book about the fivesenses, go on a sensor walk

Session Sensing andintroductionto repeats

Review what is a sensor, program arobot to dance to “If You areHappy and You Know it”

Talk about the meaning of the word“repeat,” review KIBO’scommands with KIBO Bingo

Session Repeat loopswithnumbers

Review sound sensor, programrobot to travel along differentmaps using repeats

Learn about dances from aroundthe world through watchingdifferent videos, createdecorations for the robot as wellas a stage area

Session Final projects Create a dance for the KIBO robotbased on a dance from aroundthe world (as a whole class)

Create sturdy decorations for therobot and a stage area, plan therobot’s dance

guests such as the principal and other administrators in order to celebrate the endof the unit.

Assessment

After curriculum implementationwas complete, the Solve-It assessmentwas admin-istered to students to assess their programming knowledge. The assessment com-bines KIBO’s programming language and playful stories to evaluate children’s mas-tery of different programming concepts. Because children worked in small groupsduring the curricular activities, it was important to implement individual assess-ments to see what types of tasks children could solve on their own.

The Solve-It assessment was developed to target areas of foundational program-ming ability and basic sequencing skills. These tasks capture student mastery ofprogramming concepts, from basic sequencing up through repeat loops. The assess-ment was verbally administered one-on-one to each student by one of the volunteerswho taught the robotics curriculum. The assessment required children to listen toa series of stories being read aloud to them about a robot. Then, children attemptedto create the robot’s program using paper versions of the KIBO programming iconsprovided for them (see Figure 9 for a student example and Table 2 for the storyprompts). Four Solve-It tasks were administered to address the following concepts:Easy Sequencing, Hard Sequencing, Easy “Wait for” Command, and Easy RepeatLoops with Number Parameters. Tasks were called easy or hard based on howmanycommands children needed to sequence (i.e., easy tasks had fewer blocks for chil-dren to sequence than hard tasks).

COMPUTERS IN THE SCHOOLS 179

Table . Solve-It story prompts and correct answers.

Solve-It number Story prompt Correct answer

Solve-It (EasySequencing)

“This story is about a robot that is a car. Have you everheard a car honk its horn? First, I want my car robot toturn on. Next, I want the car robot to honk thehorn—Beep! Beep!—to warn people that it’s about tomove. Then I want my car to drive straight ahead, andthen stop. So in this story, my robot turns on, beeps,goes forward, and then stops. Can you make a programthat matches this story?”

Begin, Beep, Forward,End

Solve-It (HardSequencing)

“This story is about a robot that drives into a puddle. I wantyou to make a program that lets my robot dry itself offafter it accidentally moves into a puddle. First, my robotwill turn on, and then it will move straight ahead—butOOPS! My robot is in a puddle! It’s going to make anoise—Beep!— as if it is saying ‘Oh no!’ Then, I want therobot to shake itself dry—shake!—and finally, turn off!So my robot will turn on, go straight ahead, beep, shake,and then stop. Can you make a program that matchesthis story?”

Begin, Forward, Beep,Shake, End

Solve-It (Easy “Wait for”Command)

“This story is about a dancing robot. This robot is in adance competition. The robot has stage fright though,so it is going to wait to start dancing until it hears a clapfrom the audience. Once it hears a clap, it will shake andkeep shaking until the end of the song. Then it will stop.So, for this story, my robot will turn on, wait to hear aclap, then shake, and then stop. Can you make aprogram that matches this story?”

Begin, Wait for Clap,Shake, End

Solve-It (Easy Repeat) “In this story, my robot is going to sleep. I want my robot tosay goodnight to everyone in the house. It has a brother,a sister, and a mommy, so it will say goodnight to threepeople. First, I want my robot to turn on. Next, I want therobot to make a noise—Beep!—where it is telling us‘Goodnight!’ I want the robot to say goodnight to threepeople, so it has to beep three times. Then, I want therobot to stop beeping, and last, to turn off.”So my robotwill turn on, repeat the beep sound three times, stopbeeping, and then stop. Can you make a program thatmatches this story?”

Begin, Repeat (),Beep, End Repeat,End

Note. Solve-It tasks are numbered based on overall difficulty of concept. For example, both the “easy”and “hard”sequencing tasks (tasks and ) are typically easier for children than the easiest “repeats” task (task ). Within eachcategory, there may be easy and hard tasks (for example, an “easy sequencing”and a “hard sequencing” task). Bothtarget the same conceptual understanding, but the more difficult task has more actions to sequence.

Each of the four Solve-It tasks described was scored on a 0–6 rubric based onhow close the children’s program came to being completely correct (a score of 6).The scoring rubric was developed and piloted by the DevTech Research Group(Strawhacker & Bers, 2015; Strawhacker, Sullivan, & Bers, 2013; Sullivan & Bers,2015).

Each question received two sub-scores based on separate criteria, including place-ment of Begin and End blocks (worth up to 3 points) and relative order of actionblocks (worth up to 3 points). The scoring rubric was developed after a pilot assess-ment was administered to identify incorrect answer patterns that could demonstratedevelopmental level rather than programming comprehension. Inter-scorer reliabil-ity tests during the development of the assessment showed precise agreement (twoitems; K = 0.902, p < 0.001; Strawhacker & Bers, 2015).

180 M. ELKIN ET AL.

Figure . Sample Solve-It task.

Results

For all tasks on the Solve-It assessment, basic descriptive statistics were calculated.On average, the children in this study were highly successful at mastering basic pro-gramming concepts after completing the curriculum. After children’s Solve-It datawere examined for general trends and coded for the types of mistakes, the data weredivided into two groups in order to compare programming performance betweenyounger and older preschoolers. Detailed analysis is presented in the following sec-tions.

Solve-It errors

Solve-It tasks were analyzed to look at students’ knowledge of various KIBO pro-gramming concepts and the types of errors students made. By looking at the typesof mistakes, we can better understand how to create developmentally appropriatecurricula that provide children the opportunity tomaster sequencing and other pro-gramming concepts.

Sixty-four (64) students elected to participate in the Solve-It tasks (if a studentwas asked to participate and he or she answered no, there was no force to completethe activity), but only 61 students completed all four tasks. Each Solve-It was scoredon a scale from 0–6, with 3 points awarded for the placement of the Begin and Endblocks, and another 3 points awarded for the relative order of the action blocks. Forthis analysis, overall scores as well as sub-scores were analyzed to look at the typesof mistakes that students made on each of the Solve-Its.

Children made a variety of different types of mistakes on the Solve-Its—someof which were syntactical (i.e., the program had a logical programming error andwould not work on a real robot) and some of which were story related (i.e., the pro-gram made syntactical sense but did not match the sequence of the story). Asidefrom Solve-It 4, most students were able to create syntactically correct programs,

COMPUTERS IN THE SCHOOLS 181

Figure . Percentage correct on each Solve-It task.

even if they did not match the story that they heard (See Figure 10). On three ofthe four Solve-Its, more than 65% of the children correctly placed Begin and Endblocks.

For Solve-It 1 (Easy Sequencing), 70.4% of students made a syntactically correctprogram (with 51.6% of students creating a syntactically correct program that alsomatched the story). Interestingly, 14% of students sequenced the action instructionscorrectly but misplaced the Begin and/or End block. The average score, on a scalefrom 0–6, was 4.67 (SD = 1.653), with students most frequently scoring 6. The sec-ond most frequent score was 4 (26.6%), and only one student received a score of 0,which indicates that he or she not onlymisplaced the Begin and End blocks, but alsodid not order the action instructions correctly.

For Solve-It 2 (Hard Sequencing), a slightly lower percentage of students (67.8%)created a syntactically correct program, with 40.3% also making a program thatmatched the story. The average total score was 4.08 (SD = 2.043), with studentsmost frequently scoring 6. Seven students (11.3%) received a score of 0.

Solve-It 3 (Easy “Wait for” Command) had the highest percentage (80.3%) of stu-dents who created a syntactically correct program, as well as the highest percentage(67.2%) of students whose programs also matched the story. The percentage of stu-dents who misplaced the Begin and/or End blocks was 13.1; also 13.1% of studentsswapped the Wait for Clap and Shake instructions. The average total score was 4.84(SD = 1.827), with students most frequently scoring a perfect score of 6. Three stu-dents received a total score of 0.

The lowest scores were seen on Solve-It 4 (Easy Repeat). Only 24.6% of studentscreated a functional program that also matched the story. The most frequent mis-take (16.4%) observed was an empty repeat loop, where no action block was placedbetween the Repeat and End Repeat blocks. Other mistakes included swapping theEnd and End Repeat blocks, as well as placing the incorrect action block inside the

182 M. ELKIN ET AL.

Table . Performance on Solve-It programming tasks.

Solve-It number Concept addressed Mean score (out of a maximum of )

Easy sequencing . (SD= .) Hard sequencing . (SD= .) Easy “wait-for” command . (SD= .) Easy repeats with numbers . (SD= .)

repeat loop. The average total score was 3.72 (SD = 1.845), with 3 being the mostcommon score. Six students received a score of 0.

Overall, the preschool students were successful in their performance on theSolve-It tasks, particularly the sequencing tasks that did not involve repeat loops.Each Solve-It was scored on a scale from 0 to 6, and the average score for Solve-Its1–3 was above 4 (See Table 3). On all Solve-Its, students scored on average higher ontheir Begin/End sub-score than their Sequence/Repeat sub-score. Of the four tasks,students performed best on Solve-Its 1 and 3, which required students to sequencefour instructions. They performed worse on Solve-It 2 (Hard Sequencing), whichrequired the sequencing of five instructions, and even lower on Solve-It 4 (EasyRepeat), which required the sequencing of seven instructions.

Solve-Its by age

A one-way independent samples t test was performed to determine if there weresignificant differences between older and younger children’s mean scores on each ofthe Solve-It tasks. Group placement was determined by a median split (median =4.91). For this analysis, we had 59 (N = 59) students because five students didnot provide their birthdays on the consent forms; 29 students were placed in the“younger” group and 30 students were placed in the “older” group. On average, theolder children performed better on all Solve-It tasks. Statistically significant differ-ences between the two groups were found on the easy (t(57) = −2.030, p < .05).Cohen’s effect size value (d = −0.54) suggested a moderate level of practical signif-icance. Statistically significant differences between the two groups were also foundon the hard sequencing tasks (t(55) = −2.813, p < .05). Further, Cohen’s effect sizevalue (d = −0.76) suggested a moderate to high practical significance. There wereno significant differences found between the groups on the Repeat and “Wait for”command tasks (See Table 4) indicating that both groups demonstrated the samemastery of Repeats and “Wait for” programming concepts.

Table . Differences between younger and older preschoolers’ Solve-It scores.

Solve-It Task Mean younger Mean older t df p Value

Solve-It (Easy Sequencing) . (SD= .) . (SD= .) −. p= .∗Solve-It (Hard Sequencing) . (SD= .) . (SD= .) − . p= .∗Solve-It (Easy “Wait-for”Command) . (SD= .) . V(SD= .) − . p= .Solve-It (Easy Repeat Loops) . (SD= .) . (SD= .) − . p= .

∗ Significant p value less than ..

COMPUTERS IN THE SCHOOLS 183

Discussion

The results from this study suggest that the KIBO robot and some aspects of theKIBO programming language are appropriate for children as young as age 3, despitethe fact that it was designed for children ages 4 and older. The introductory roboticscurriculum used in this study focused on rudimentary programming skills forKIBO, including sequencing and an introduction to repeat loops. Preschool chil-dren in the study, ages 3 to 5, were able to successfully master sequencing a syn-tactically correct program. However, the more instructions the students were askedto sequence, the more difficult it was for them to correctly create a program. Thisis consistent with the findings of Sullivan and Bers (2015), who found that pre-kindergarten students were more successful on an easy-sequencing Solve-It task(ordering four blocks) than a hard-sequencing Solve-It task (ordering five blocks).This is also consistent with the literature showing that younger children do not haveenough working memory to hold five instructions simultaneously in their mindsuntil they are several years older (Shonkoff et al., 2011).

The students’ success on sequencing shorter programs may be due to their work-ing memory and the capacity to remember all the parts of a longer story at a giventime.Workingmemory is described as the ability to simultaneously hold andmanip-ulate information internally over a short period of time (Shonkoff et al., 2011). Dur-ing the Solve-It tasks, students had to simultaneously process the story being told,remember the programming instructions they had learned, and connect the instruc-tions to the story. All of these elements in their mind may have been too heavy ofa cognitive load for the young children in this study, even if the programming con-cepts were manageable.

Regardless of age, students in this study scored much lower on Solve-It 4. Thissuggests that programming repeat loops may be a challenging concept for veryyoung children, either due to the number of blocks needed or their conceptualunderstanding of the repeat loop. The types of mistakes that students made on thisSolve-It suggest that most children misunderstood the syntactical rules of creatingrepeat loops. Additionally, many students swapped the positions of the End and EndRepeat blocks, suggesting that students had trouble distinguishing between the Endblock (which ends thewhole program) and EndRepeat block (which ends the repeatloop). Physically, both blocks include theword “end” and use the color red.However,when manipulating the blocks, the End and End Repeat blocks have different phys-ical features (such as the absence of a peg on the End block); these features were notpresent when using the paper version of the blocks for the Solve-It assessment. Thismay have added to the difficulty for students to fully differentiate between the blockswhen they needed to be used in one program. Future work may want to include ablock-identification task to better understand if these young preschoolers can iden-tify different blocks, and then using the Solve-It assessment to see if they can applywhat they know about the blocks to create a syntactically correct program.

Given that less than a quarter of the children were able to create a functionalprogram that included repeats, it may be worthwhile when introducing suchyoung children to programming to focus on basic sequencing, or to provide more

184 M. ELKIN ET AL.

scaffolding when teaching repeat loops. Repeat loops involve more than just newblocks; they also introduce a new piece of KIBO syntax (i.e., creating a “parenthe-sis” with the Repeat and End Repeat block to separate a series of commands fromthe rest of the program). In addition to holding this new piece of syntax in theirworkingmemory, repeat loops also require children to estimate andmathematicallyreason with number parameters. This may be too much of a cognitive strain forchildren beginning to program. This connects with previous findings that preschoolstudents spend more time than kindergarten through second-grade students onbasic robotics concepts, and move through an introductory robotics curriculum ata slower pace (Sullivan & Bers, 2015; Sullivan et al. 2013).

Results also indicate that older preschoolers (around 5 years old) demonstrateda higher mean level of mastery on all programming concepts assessed than youngerpreschoolers (under 5 years). This may be due to a variety of factors includingincreased working memory, attention span, and ability to plan (Shonkoff et al.,2011). These results suggest that, even in a preschool setting, teachers should con-sider offering differentiated learning opportunities for students based on cognitiveand social development. Additionally, when givenmore time, teachersmay find thatpreschoolers, particularly the older students, may be able to master more program-ming concepts beyond those introduced in this study. Because KIBO programmingconcepts build on one another, children can easily continue to explore and mas-ter more complicated programming concepts including repeat loops with sensorparameters and conditional branching. This makes the kit ideal for preschool set-tings with a range of student abilities.

Limitations and recommendations for future research

The primary limitation of this study was the availability of robotic materials at theschool. In order to keep the ratio of three children to one robot, robotics kits neededto be shared between classrooms. As a result, of the 9 hours in which students partic-ipated in the robotics curriculum, only half the time was spent with the robot itself.While the non-robotics time was a great opportunity for students to play games andread stories reinforcing what they learned during robotics time, students did nothave a lot of time to engage in hands-on programming. Given that the assessmentmeasured programming ability, students may have done better if they had receivedmore practice with the curriculum.

Another limitation of this study correlates with the scheduling and logistics ofworking with students of all ages. Due to the schedules of both the preschoolersand the student volunteers from Tufts University, the six sessions were spread outwith lengthy gaps between each one. Each lesson built off of the previous sessions,so students were asked to recall information that they had been taught a while ago.If implemented again, a longer intervention with more consistent timing may behelpful.

Furthermore, the way in which the Solve-It assessments were administered mayhave hindered student performance. The curriculum allowed time for students tocreate programs for their robot using tangible blocks, scan the program onto the

COMPUTERS IN THE SCHOOLS 185

robot, and then see the robot move. In this way, students were able to directly checkif the program they had created made the robot move like they had intended itto. Additionally, the curriculum was designed to be open-ended, so children couldchoose whichever instructions they wanted for their robot to act out. Conversely,on the Solve-It assessment, students were given a specific story that they neededto recreate using paper representations of the blocks. Students did not have theopportunity to see the robot act out the program and decide whether they neededto change any instructions. Future studies may want to develop an assessment thatgives students the opportunity to use the tangible blocks and check their robot’s pro-gram for each story.

Conclusion

Robotics offers preschool children and teachers a playful new way to learn founda-tional engineering and programming concepts. This study demonstrated that it ispossible to teach preschool children as young as 3 years of age fundamental pro-gramming concepts such as sequencing and repeat loops. As results from this studyshow, with proper scaffolding and time to explore engineering concepts throughrobotic and non-robotic activities, students as young as 3 can successfully build andprogram a KIBO robot.

References

American Academy of Pediatrics. (2003). Prevention of pediatric overweight and obesity: Policystatement. Pediatrics, 112, 424–430.

Bers, M. (2008). Blocks to robots :Learning with technology in the early childhood classroom. NewYork, NY: Teachers College Press.

Bers, M. U. (2012). Designing digital experiences for positive youth development: From playpen toplayground. Cary, NC: Oxford.

Bers, M. U., Ponte, I., Juelich, K., Viera, A., & Schenker, J. (2002). Teachers as designers: Integrat-ing robotics into early childhood education. Information Technology in Childhood Education,1, 123–145.

Cejka, E., Rogers, C., & Portsmore, M. (2006). Kindergarten robotics: Using robotics to moti-vate math, science, and engineering literacy in elementary school. International Journal ofEngineering Education, 22(4), 711–722.

Clements, D. H. (1999). Young children and technology. In G. D. Nelson (Ed.), Dialogue on earlychildhood science, mathematics, and technology education. Washington, DC: American Asso-ciation for the Advancement of Science.

Common SenseMedia. (2013).Zero to eight: Children’s media use in America 2013. San Francisco,CA: Author.

Druin, A., & Hendler, J. (Eds.). (2000). Robots for kids: Exploring new technologies for learningexperiences. San Francisco, CA: Morgan Kaufman/Academic Press.

Fessakis, G., Gouli, E., & Mavroudi, E. (2013). Problem solving by 5–6 sources old kindergartenchildren in a computer programming environment: A case study. Computers & Education,63, 87–97.

186 M. ELKIN ET AL.

Kazakoff, E., Sullivan, A., & Bers, M. U. (2013). The effect of a classroom-based intensive roboticsand programming workshop on sequencing ability in early childhood. Early Childhood Edu-cation Journal, 41(4), 245–255. doi:10.1007/s10643-012-0554-5.

National Association for the Education of Young Children (NAEYC), & FredRogers Center (2012). Technology and interactive media as tools in early child-hood programs serving children from birth through age 8. Retrieved fromhttp://www.naeyc.org/files/naeyc/file/positions/PS_technology_WEB2.pdf

Kazakoff, E. R., & Bers, M. U. (2011, April). The impact of computer programming on sequencingability in early childhood. Paper presented at the American Educational Research AssociationConference (AERA), New Orleans, LA.

Lee, K., Sullivan, A., Bers, M. U. (2013). Collaboration by design: Using robotics to foster socialinteraction in kindergarten. Computers in the Schools, 30(3), 271–281.

Office of Educational Technology. (2010). Transforming American education: Learning pow-ered by technology. Washington, DC: U.S. Department of Education. Retrieved fromhttp://www.ed.gov/technology/netp-2010

Peel, S., & Prinsloo, F. (2001). Neuro- integration movements workshop: Module 1. South Africa:Brain Gym Edu-K (Educational Kinesiology).

Perlman, R. (1976). Using computer technology to provide a creative learning environment forpreschool children [Logo memo no. 24]. Cambridge, MA: MIT Artificial Intelligence Labo-ratory Publications 260.

Piaget, J. (1952). The origins of intelligence in children. New York, NY: International UniversitiesPress.

Rhode Island Department of Education (RIDE). (2013). Rhode Island early learningand development standards. Retrieved from http://www.ride.ri.gov/InstructionAsses-sment/EarlyChildhoodEducation/EarlyLearningandDevelopmentStandards.aspx#1669797-literacy-l

Shonkoff, J. P., Duncan, G. J., Fisher, P. A., Magnuson, K., & Raver, C. (2011). Building the brain’s“air traffic control” system: How early experiences shape the development of executive function(Working Paper No. 11). Retrieved from http://www.developingchild.harvard.edu

Strawhacker, A., Sullivan, A., & Bers, M.U. (2013). TUI, GUI, HUI: Is a bimodal interface trulyworth the sum of its parts? Proceedings of the 12th International Conference on InteractionDesign and Children (IDC ′′13; pp. 309–312). New York, NY: Association for ComputingMachinery (ACM).

Strawhacker, A. L., & Bers,M.U. (2015). “I wantmy robot to look for food”: Comparing children’sprogramming comprehension using tangible, graphical, and hybrid user interfaces. Interna-tional Journal of Technology and Design Education, 25(3), 293–319.

Sullivan, A., & Bers, M. U. (2015). Robotics in the early childhood classroom: Learning outcomesfrom an eight-week robotics curriculum in pre-kindergarten through second grade. Interna-tional Journal of Technology and Design Education, 36(1), 3–20.

Sullivan, A., Elkin, M., & Bers, M. U. (2015). KIBO robot demo: Engaging young children in pro-gramming and engineering. Proceedings of the 14th International Conference on Interac-tion Design and Children (IDC ′′15; pp. 418–421). Boston, MA: Association for ComputingMachinery.

Sullivan, A., Kazakoff, E. R., & Bers, M.U. (2013). The wheels on the Bot go round and round:Robotics curriculum in pre-kindergarten. Journal of Information Technology Education: Inno-vations in Practice, 12, 203–219.

Svensson,A. K. (2000). Computers in school: Socially isolating or a tool to promote collaboration?Journal of Educational Computing Research, 22(4), 437–453.

Wyeth, P. (2008). How young children learn to program with sensor, action, and logic blocks.International Journal of the Learning Sciences, 17(4), 517–550.

Zelazo, P. D., Carter, A., Reznick, J. S., & Frye, D. (1997). Early development of executive function:A problem-solving framework. Review of General Psychology, 1(2), 198–226.